3.4 cD Galaxies - Case Western Reserve University

17
stoppmg mechanism. 3.4 cD Galaxies The term cD galaxy was introduced by Matthews, Morgan & Schmidt (1964) to describe a galaxy with the nucleus of a giant elliptical surrounded by an extended, slowly decreasing envelope. The cD galaxies are of large extent (they are the largest known galaxies) and high luminosity, and frequently contain multiple nuclei. Several lists of Abell clusters that contain cD galaxies have been published (e.g. Matthews, Morgan & Schmidt 1964, Morgan & Lesh 1965; see also Bautz & Morgan 1970, Bautz 1972, Rood & Sastry 1971, Leir & van den Bergh 1977). Approximately 20% of rich clusters of galaxies contain a dominant central cD galaxy. Recently Morgan, Kayser & White (1975)"and Albert, White & Morgan (1977) investigated the possible existence of cD galaxies in poor clusters and· listed suspected cD galaxies in some small groups (cf also van den Bergh 1975). Since all previous examples of cD's were located in rich clusters, the newly reported objects are important if they are really cD's and if they are valid members of the 19 20 a 21 . 22 23 24 SJ 25 26 27 28 29 30 0.5 1.0 1.5 2.0 2.5 3.0 0.0 log r (kpc) Figure 1 Surface brightness profile of the cD galaxy in A2670 (measured by Oemler 1973a). SJ is in mag (arc sec)-Z; open circles are green (J) magnitudes and filled circles are red magnitudes shifted by + 1.1 mag. The solid line is the profile of a normal elliptical galaxy with a length scale a. The dashed line "represents the relation a(r) ex: r- 1 6 suggested by Equation 4 (Section 3.4).

Transcript of 3.4 cD Galaxies - Case Western Reserve University

Page 1: 3.4 cD Galaxies - Case Western Reserve University

stoppmg mechanism

34 cD Galaxies

The term cD galaxy was introduced by Matthews Morgan amp Schmidt (1964) to describe a galaxy with the nucleus of a giant elliptical surrounded by an extended slowly decreasing envelope The cD galaxies are of large extent (they are the largest known galaxies) and high luminosity and frequently contain multiple nuclei Several lists of Abell clusters that contain cD galaxies have been published (eg Matthews Morgan amp Schmidt 1964 Morgan amp Lesh 1965 see also Bautz amp Morgan 1970 Bautz 1972 Rood amp Sastry 1971 Leir amp van den Bergh 1977) Approximately 20 of rich clusters of galaxies contain a dominant central cD galaxy Recently Morgan Kayser amp White (1975)and Albert White amp Morgan (1977) investigated the possible existence of cD galaxies in poor clusters andmiddot listed suspected cD galaxies in some small groups (cf also van den Bergh 1975) Since all previous examples of cDs were located in rich clusters the newly reported objects are important if they are really cDs and if they are valid members of the

19

bull20 a

21

22

23

24 SJ

25

26

27

28

29

3005 10 15 20 25 3000

log r (kpc)

Figure 1 Surface brightness profile of the cD galaxy in A2670 (measured by Oemler 1973a) SJ is in mag (arc sec)-Z open circles are green (J) magnitudes and filled circles are red magnitudes shifted by + 11 mag The solid line is the profile of a normal elliptical galaxy with a length scale a The dashed line represents the relation a(r) ex r- 1bull6 suggested by Equation 4 (Section 34)

ampvampr 990 ~ fy1lJ~gt~ of~ eeL~

CU It ~ etttdltabW s~ 3~C

~ ~ ~~ tgt~~s ~ ~

k Y ko-oltgtS~ ~~

oO~ to frr orcb ~

Q

~~ - 0 C )

04t lt (j ~

gt ~Q gt0

bull bull

bull bull bullbullbull bull

bullbull bull bull bull

26

h~~ ~~

-IN 24-3 -J 22 01

Q

20

I I I 26 - -f---4 bull shy

-Nbull-

bullbull bull bull

bullbull -I 24

3- bull I shybull -0 gt- bull bull bullbullbull - -l 22-

01bull-r --bull-~)- -- - - -T Q( _ bull bull

o bull bull _ - 20 I I I

95 10 105 11 115

42 --()-01 41 shy

QI

)( --l

01 40 Q

39

log LB solar

bull 42 bullbullo---f--o bull

()

bull middot1 bullbull - shy

-~41

-QI

x

bullbull

-l lt1-40 01

Q

39 95 10 105 11 115

log LB solar

I

bull T I

+bull Ibull I bull I bullbullbull bull bull bull I

bull bull I bullbull bull bullbull I bull I- -i I I --bullbull - - - - - - - - - I

TI bull bull bull bull bullbull middota bull bullbull I bull bullbull

-1 0 1 2 3 0(4)0)(100

bull I bullbull I bullbull I + - ~

I I

bull bull I

IIbull I bull I

I

I I

-1 0 1 2 3 0(4)0)(100

Figure 5 The radio and X-ray properties of elliptical galaxies against blue luminosity LB andshyisophote shape a4a In the tiLx against 0-4a diagram the contribution of the discrete sources to the total X-ray emission (indicated by the line in the Lx against LB plot) is subtracted according to Canizares et al (1987) LR gives the total radio luminosity at 14 GHz values below 1021 WHz are nearly all upper limits

(I) Q~ ~~~~ (~ pound~ ~ $r

ak~di~~~~

x

lIow i7lt-you H1Ji- r(e ~aUshybull

JJoub ~ ~ e+~ ~ ~tL

Y O~ ueL~ so~ bull ~

~ ~ ~~~ ~rcAs middot

v~ middot ~~~ ~ 6if ~ ~fy~~ ~~~

bullbullbull

10 - w ~-y~ JaI~ ~ sMs

-s~ ~ ) ~ ~St7

4Ii7~~~sectefJf-f~ ~~~~~wd

J ~ of ~~~ ~ ~ ~ds ~ amp4 k ~ ~ ~ +

2J ~s~

PLATE 4

I

It

~

~

~ ~middot~~gti~H- 8~ ~~ ~~ _I- f

~~~ bull bull --00gt010004

bull ~

~IiitI~--~~~ ~ ~ ~

middoti ~~~ shy j ~bull

~ ~1 -

I I

FIG 4-NGC 3923 The top picture (Fig 4a) was made from thrce UK Schmidt I1Ia-J plates enhanced to show the NE shell Scale bar ~ is 10 The lower pictures (Figs 4h and 4() were mac from the same AAT plate printed through un~harp masks with diff~rent characteristics About l~ shells can be counted The dust cloud in Fig 4( is cal Scale bars are 5 and 2 in Fig 4h and 4c respecllvcI

M LI AND CARTER (see page 539)

06

_----_- -----shy

~

~ ~

~ ~ ~

bull

tl r-

MALIN AND CARTER (see page 538) ~

0 D

ltl

N

~ exgt

M2 6

10 14 18

~~

pound~

FIG I-Time evolution of the radial planar encounter between an exponential surface density disk and a 10 times more massive fixed and rigid isochrone potential The cross denotes the center of the isochrone potential and the disk is initially moving to the left The evolution is viewed from above the orbit plane (X -Y plane) and the times indicated are in units of the circular period at a radius of one scale length in the isochrone potential The bar under the initial time is 10 scale lengths long The disk as seen in the figure is rotating in a counterclockwise sense

2

10 14 18

FIG 2-Time evolution of the radia~ planar encounter of an exponential surface density disk and a 100 times more massive fixed and rigid isochrone potential Other figure parameters are as in Fig I

598

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

o D

ltl 602 rshyN

3 ~ co

- 0 ~ -1

-3

QUINN Vol 279

I I I

- ~

-

- ~

~

-~ii~~1~fllt-lt ~ - _ shy

- ~-~ bull

~

I I I I

o 5 10 15 20 radius

FIG 6-The radial velocity-radius plane for particles in the model shown iii Fig 2 at time 18 Radii are in units of the isochrone scale length and velocities are in units of (GM10a)1I2

dimensional system of test particles falling from rest into a fixed potential The extent of the system in the field introduces a spread of energies across the system For potentials other than the simple harmonic potential the range in energies corresshyponds to a range in periods with the most bound particles having the shortest periods Figure 5 shows the phase evolution of such a system As time passes the shorter period particles begin to lead the longer period ones and the system starts to wrap in the phase plane The wrapping proceeds at a rate determined by the range in periods present and the number of wraps at time t after the infall begins is simply

(1)

where Q is the radial frequency andQax aIld Qin are the maximum and minimum frequencies present respectively The spatial evolution of the system can be found by projecting the phase plot onto the spatial coordinate axis The maximum spatial excursion of each phase wrap corresponds to a sharply defined density maximum The density maxima occur near the turnaround points of the particle orbits and propagate slowly in radius to the outermost turning point set by the least bound particle The density maxima are therefore propagating density waves and are similar to the ring structures produced by Lynds and Toomre (1976) in their ring galaxy simulations A similar dynamical picture has been proposed for the evolution of large-scale structure in the universe (Zeldovich 1970 Doroshkevich et at 1980)

Figure 6 shows the radial velocityradius plane for the model presented in Figure 2 Here the phase wrapping nature of the shells can be clearly seen Note also that a much larger number of shells can be detected in this way than fromjust the particle distribution This is because the small number of particles used

is incapable of showing all the shells at a sufficiently high contrast to be seen

The phase wrapping interpretation of shell structures has a number of desirable features First the phase wraps are interleaved in radius as has been observed for the shell galaxies Second the range of the number of shells present around ellipticals is a simple consequence of the age of the event More shells will imply that a longer time has passed since the merger event given similar rates of shells production Shell producshytion rates can be estimated from the range in radii of the shells and hence the spread in periods present

The position in space of a particular shell can be calculated for a given potential by appreciating that the shells occur close to the radial turning points of the orbits Hence

Qdm t 2n(m - em) (2)

where Qd~ is the radial frequency ora particle with its radial turning point at radius dm dm is the distance of the shell from the center of the potential em is the orbital phase of particles with turning points at dm and m labels the shells by the number of orbital periods completed at time t by particles in the shell at that time The orbital phase of a particle is defined to be the time required to travel around its orbit to the nearest turning point in units of the orbital period hence -05s em s 05 If all the particles begin from rest then em = 0 for all particles and Qd

m(t2n) is an integer for all shells The variable m is the

proper shell number in that it is the total number of completed periods A second number n the observed number of shells is defined to be

n=m-t+1 (3)

where t is the total number of completed periods for particles in the outermost shell The outermost shell is therefore labeled

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

43 Photometry of Elliptical Galaxies 203

~ 095

I

SshyI

S 09

Figure 441 Galaxies with fine structure have bluer colors Here we plot the correlation between the fine-structure parameter ~ [equashytion (437)] and the color (B-V)-21

085 This is the galaxys color after corshyrection for the color-magnitude efshyfect (sect434) to absolute magnitude

08 0 2 4 6 8

MB = -21 [After Schweizer amp Seitzer (1992) from data kindly sup-

L plied by F Schweizer]

very hard to detect in late-type systems so we do not know how universal a phenomenon they are

A stellar system can display a sharp edge only if some parts of its phase space (see BT sect41) are very much more densely populated with stars than neighboring parts of phase space In the classical dynamical model of an elliptical (BT sect44) phase space is populated very smoothly Therefore the existence of ripples directly challenges the classical picture of ellipticals One likely possibility is that ellipticals acquire ripples late in life as a result of accreting material from a system within which there are relatively large grashydients in phase-space density Systems with large density gradients in phase space include disk galaxies and dwarf galaxies in a thin disk the phase-space density of stars peaks strongly around the locations of circular orbits while in a dwarf galaxy all stars move at approximately the systemic velocity so that there is only a small spread in velocity space

Numerical simulations suggest that ripples can indeed form when mateshyrial is accreted from either a disk galaxy or a dwarf system - see Barnes amp Hernquist (1992) for a review Moreover simulations have successfully reshyproduced the interleaved property of ripples described above Despite these successes significant uncertainties still surround the ripple phenomenon beshycause the available simulations have important limitations and it is not clear how probable their initial conditions are

Schweizer et al (1990) defined an index E that quantifies the amount of fine structure such as ripples that a galaxy possesses

E == S + 10g(1 + n) + J + B + X (437)

Here S measures the strength of the most prominent ripple on a scale of 0 to 3 n is the number of detected ripples J is the number of optical jets B is a measure of the boxiness of the galaxys isophotes on a scale of 0 to 4 X is o or 1 depending on whether the galaxys image shows an X structure For a sample of 69 nearby early-type galaxies E varies from 0 to 76 Notice that

VJe IJ- TIIEOIltpoundlJ

~r~d ampk~ 0 ~~1ud)~

0U- ccdd we M-la)W ~ wocdd J~ ~o~~

c- -$2 r

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 2: 3.4 cD Galaxies - Case Western Reserve University

ampvampr 990 ~ fy1lJ~gt~ of~ eeL~

CU It ~ etttdltabW s~ 3~C

~ ~ ~~ tgt~~s ~ ~

k Y ko-oltgtS~ ~~

oO~ to frr orcb ~

Q

~~ - 0 C )

04t lt (j ~

gt ~Q gt0

bull bull

bull bull bullbullbull bull

bullbull bull bull bull

26

h~~ ~~

-IN 24-3 -J 22 01

Q

20

I I I 26 - -f---4 bull shy

-Nbull-

bullbull bull bull

bullbull -I 24

3- bull I shybull -0 gt- bull bull bullbullbull - -l 22-

01bull-r --bull-~)- -- - - -T Q( _ bull bull

o bull bull _ - 20 I I I

95 10 105 11 115

42 --()-01 41 shy

QI

)( --l

01 40 Q

39

log LB solar

bull 42 bullbullo---f--o bull

()

bull middot1 bullbull - shy

-~41

-QI

x

bullbull

-l lt1-40 01

Q

39 95 10 105 11 115

log LB solar

I

bull T I

+bull Ibull I bull I bullbullbull bull bull bull I

bull bull I bullbull bull bullbull I bull I- -i I I --bullbull - - - - - - - - - I

TI bull bull bull bull bullbull middota bull bullbull I bull bullbull

-1 0 1 2 3 0(4)0)(100

bull I bullbull I bullbull I + - ~

I I

bull bull I

IIbull I bull I

I

I I

-1 0 1 2 3 0(4)0)(100

Figure 5 The radio and X-ray properties of elliptical galaxies against blue luminosity LB andshyisophote shape a4a In the tiLx against 0-4a diagram the contribution of the discrete sources to the total X-ray emission (indicated by the line in the Lx against LB plot) is subtracted according to Canizares et al (1987) LR gives the total radio luminosity at 14 GHz values below 1021 WHz are nearly all upper limits

(I) Q~ ~~~~ (~ pound~ ~ $r

ak~di~~~~

x

lIow i7lt-you H1Ji- r(e ~aUshybull

JJoub ~ ~ e+~ ~ ~tL

Y O~ ueL~ so~ bull ~

~ ~ ~~~ ~rcAs middot

v~ middot ~~~ ~ 6if ~ ~fy~~ ~~~

bullbullbull

10 - w ~-y~ JaI~ ~ sMs

-s~ ~ ) ~ ~St7

4Ii7~~~sectefJf-f~ ~~~~~wd

J ~ of ~~~ ~ ~ ~ds ~ amp4 k ~ ~ ~ +

2J ~s~

PLATE 4

I

It

~

~

~ ~middot~~gti~H- 8~ ~~ ~~ _I- f

~~~ bull bull --00gt010004

bull ~

~IiitI~--~~~ ~ ~ ~

middoti ~~~ shy j ~bull

~ ~1 -

I I

FIG 4-NGC 3923 The top picture (Fig 4a) was made from thrce UK Schmidt I1Ia-J plates enhanced to show the NE shell Scale bar ~ is 10 The lower pictures (Figs 4h and 4() were mac from the same AAT plate printed through un~harp masks with diff~rent characteristics About l~ shells can be counted The dust cloud in Fig 4( is cal Scale bars are 5 and 2 in Fig 4h and 4c respecllvcI

M LI AND CARTER (see page 539)

06

_----_- -----shy

~

~ ~

~ ~ ~

bull

tl r-

MALIN AND CARTER (see page 538) ~

0 D

ltl

N

~ exgt

M2 6

10 14 18

~~

pound~

FIG I-Time evolution of the radial planar encounter between an exponential surface density disk and a 10 times more massive fixed and rigid isochrone potential The cross denotes the center of the isochrone potential and the disk is initially moving to the left The evolution is viewed from above the orbit plane (X -Y plane) and the times indicated are in units of the circular period at a radius of one scale length in the isochrone potential The bar under the initial time is 10 scale lengths long The disk as seen in the figure is rotating in a counterclockwise sense

2

10 14 18

FIG 2-Time evolution of the radia~ planar encounter of an exponential surface density disk and a 100 times more massive fixed and rigid isochrone potential Other figure parameters are as in Fig I

598

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

o D

ltl 602 rshyN

3 ~ co

- 0 ~ -1

-3

QUINN Vol 279

I I I

- ~

-

- ~

~

-~ii~~1~fllt-lt ~ - _ shy

- ~-~ bull

~

I I I I

o 5 10 15 20 radius

FIG 6-The radial velocity-radius plane for particles in the model shown iii Fig 2 at time 18 Radii are in units of the isochrone scale length and velocities are in units of (GM10a)1I2

dimensional system of test particles falling from rest into a fixed potential The extent of the system in the field introduces a spread of energies across the system For potentials other than the simple harmonic potential the range in energies corresshyponds to a range in periods with the most bound particles having the shortest periods Figure 5 shows the phase evolution of such a system As time passes the shorter period particles begin to lead the longer period ones and the system starts to wrap in the phase plane The wrapping proceeds at a rate determined by the range in periods present and the number of wraps at time t after the infall begins is simply

(1)

where Q is the radial frequency andQax aIld Qin are the maximum and minimum frequencies present respectively The spatial evolution of the system can be found by projecting the phase plot onto the spatial coordinate axis The maximum spatial excursion of each phase wrap corresponds to a sharply defined density maximum The density maxima occur near the turnaround points of the particle orbits and propagate slowly in radius to the outermost turning point set by the least bound particle The density maxima are therefore propagating density waves and are similar to the ring structures produced by Lynds and Toomre (1976) in their ring galaxy simulations A similar dynamical picture has been proposed for the evolution of large-scale structure in the universe (Zeldovich 1970 Doroshkevich et at 1980)

Figure 6 shows the radial velocityradius plane for the model presented in Figure 2 Here the phase wrapping nature of the shells can be clearly seen Note also that a much larger number of shells can be detected in this way than fromjust the particle distribution This is because the small number of particles used

is incapable of showing all the shells at a sufficiently high contrast to be seen

The phase wrapping interpretation of shell structures has a number of desirable features First the phase wraps are interleaved in radius as has been observed for the shell galaxies Second the range of the number of shells present around ellipticals is a simple consequence of the age of the event More shells will imply that a longer time has passed since the merger event given similar rates of shells production Shell producshytion rates can be estimated from the range in radii of the shells and hence the spread in periods present

The position in space of a particular shell can be calculated for a given potential by appreciating that the shells occur close to the radial turning points of the orbits Hence

Qdm t 2n(m - em) (2)

where Qd~ is the radial frequency ora particle with its radial turning point at radius dm dm is the distance of the shell from the center of the potential em is the orbital phase of particles with turning points at dm and m labels the shells by the number of orbital periods completed at time t by particles in the shell at that time The orbital phase of a particle is defined to be the time required to travel around its orbit to the nearest turning point in units of the orbital period hence -05s em s 05 If all the particles begin from rest then em = 0 for all particles and Qd

m(t2n) is an integer for all shells The variable m is the

proper shell number in that it is the total number of completed periods A second number n the observed number of shells is defined to be

n=m-t+1 (3)

where t is the total number of completed periods for particles in the outermost shell The outermost shell is therefore labeled

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

43 Photometry of Elliptical Galaxies 203

~ 095

I

SshyI

S 09

Figure 441 Galaxies with fine structure have bluer colors Here we plot the correlation between the fine-structure parameter ~ [equashytion (437)] and the color (B-V)-21

085 This is the galaxys color after corshyrection for the color-magnitude efshyfect (sect434) to absolute magnitude

08 0 2 4 6 8

MB = -21 [After Schweizer amp Seitzer (1992) from data kindly sup-

L plied by F Schweizer]

very hard to detect in late-type systems so we do not know how universal a phenomenon they are

A stellar system can display a sharp edge only if some parts of its phase space (see BT sect41) are very much more densely populated with stars than neighboring parts of phase space In the classical dynamical model of an elliptical (BT sect44) phase space is populated very smoothly Therefore the existence of ripples directly challenges the classical picture of ellipticals One likely possibility is that ellipticals acquire ripples late in life as a result of accreting material from a system within which there are relatively large grashydients in phase-space density Systems with large density gradients in phase space include disk galaxies and dwarf galaxies in a thin disk the phase-space density of stars peaks strongly around the locations of circular orbits while in a dwarf galaxy all stars move at approximately the systemic velocity so that there is only a small spread in velocity space

Numerical simulations suggest that ripples can indeed form when mateshyrial is accreted from either a disk galaxy or a dwarf system - see Barnes amp Hernquist (1992) for a review Moreover simulations have successfully reshyproduced the interleaved property of ripples described above Despite these successes significant uncertainties still surround the ripple phenomenon beshycause the available simulations have important limitations and it is not clear how probable their initial conditions are

Schweizer et al (1990) defined an index E that quantifies the amount of fine structure such as ripples that a galaxy possesses

E == S + 10g(1 + n) + J + B + X (437)

Here S measures the strength of the most prominent ripple on a scale of 0 to 3 n is the number of detected ripples J is the number of optical jets B is a measure of the boxiness of the galaxys isophotes on a scale of 0 to 4 X is o or 1 depending on whether the galaxys image shows an X structure For a sample of 69 nearby early-type galaxies E varies from 0 to 76 Notice that

VJe IJ- TIIEOIltpoundlJ

~r~d ampk~ 0 ~~1ud)~

0U- ccdd we M-la)W ~ wocdd J~ ~o~~

c- -$2 r

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 3: 3.4 cD Galaxies - Case Western Reserve University

bull bull

bull bull bullbullbull bull

bullbull bull bull bull

26

h~~ ~~

-IN 24-3 -J 22 01

Q

20

I I I 26 - -f---4 bull shy

-Nbull-

bullbull bull bull

bullbull -I 24

3- bull I shybull -0 gt- bull bull bullbullbull - -l 22-

01bull-r --bull-~)- -- - - -T Q( _ bull bull

o bull bull _ - 20 I I I

95 10 105 11 115

42 --()-01 41 shy

QI

)( --l

01 40 Q

39

log LB solar

bull 42 bullbullo---f--o bull

()

bull middot1 bullbull - shy

-~41

-QI

x

bullbull

-l lt1-40 01

Q

39 95 10 105 11 115

log LB solar

I

bull T I

+bull Ibull I bull I bullbullbull bull bull bull I

bull bull I bullbull bull bullbull I bull I- -i I I --bullbull - - - - - - - - - I

TI bull bull bull bull bullbull middota bull bullbull I bull bullbull

-1 0 1 2 3 0(4)0)(100

bull I bullbull I bullbull I + - ~

I I

bull bull I

IIbull I bull I

I

I I

-1 0 1 2 3 0(4)0)(100

Figure 5 The radio and X-ray properties of elliptical galaxies against blue luminosity LB andshyisophote shape a4a In the tiLx against 0-4a diagram the contribution of the discrete sources to the total X-ray emission (indicated by the line in the Lx against LB plot) is subtracted according to Canizares et al (1987) LR gives the total radio luminosity at 14 GHz values below 1021 WHz are nearly all upper limits

(I) Q~ ~~~~ (~ pound~ ~ $r

ak~di~~~~

x

lIow i7lt-you H1Ji- r(e ~aUshybull

JJoub ~ ~ e+~ ~ ~tL

Y O~ ueL~ so~ bull ~

~ ~ ~~~ ~rcAs middot

v~ middot ~~~ ~ 6if ~ ~fy~~ ~~~

bullbullbull

10 - w ~-y~ JaI~ ~ sMs

-s~ ~ ) ~ ~St7

4Ii7~~~sectefJf-f~ ~~~~~wd

J ~ of ~~~ ~ ~ ~ds ~ amp4 k ~ ~ ~ +

2J ~s~

PLATE 4

I

It

~

~

~ ~middot~~gti~H- 8~ ~~ ~~ _I- f

~~~ bull bull --00gt010004

bull ~

~IiitI~--~~~ ~ ~ ~

middoti ~~~ shy j ~bull

~ ~1 -

I I

FIG 4-NGC 3923 The top picture (Fig 4a) was made from thrce UK Schmidt I1Ia-J plates enhanced to show the NE shell Scale bar ~ is 10 The lower pictures (Figs 4h and 4() were mac from the same AAT plate printed through un~harp masks with diff~rent characteristics About l~ shells can be counted The dust cloud in Fig 4( is cal Scale bars are 5 and 2 in Fig 4h and 4c respecllvcI

M LI AND CARTER (see page 539)

06

_----_- -----shy

~

~ ~

~ ~ ~

bull

tl r-

MALIN AND CARTER (see page 538) ~

0 D

ltl

N

~ exgt

M2 6

10 14 18

~~

pound~

FIG I-Time evolution of the radial planar encounter between an exponential surface density disk and a 10 times more massive fixed and rigid isochrone potential The cross denotes the center of the isochrone potential and the disk is initially moving to the left The evolution is viewed from above the orbit plane (X -Y plane) and the times indicated are in units of the circular period at a radius of one scale length in the isochrone potential The bar under the initial time is 10 scale lengths long The disk as seen in the figure is rotating in a counterclockwise sense

2

10 14 18

FIG 2-Time evolution of the radia~ planar encounter of an exponential surface density disk and a 100 times more massive fixed and rigid isochrone potential Other figure parameters are as in Fig I

598

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

o D

ltl 602 rshyN

3 ~ co

- 0 ~ -1

-3

QUINN Vol 279

I I I

- ~

-

- ~

~

-~ii~~1~fllt-lt ~ - _ shy

- ~-~ bull

~

I I I I

o 5 10 15 20 radius

FIG 6-The radial velocity-radius plane for particles in the model shown iii Fig 2 at time 18 Radii are in units of the isochrone scale length and velocities are in units of (GM10a)1I2

dimensional system of test particles falling from rest into a fixed potential The extent of the system in the field introduces a spread of energies across the system For potentials other than the simple harmonic potential the range in energies corresshyponds to a range in periods with the most bound particles having the shortest periods Figure 5 shows the phase evolution of such a system As time passes the shorter period particles begin to lead the longer period ones and the system starts to wrap in the phase plane The wrapping proceeds at a rate determined by the range in periods present and the number of wraps at time t after the infall begins is simply

(1)

where Q is the radial frequency andQax aIld Qin are the maximum and minimum frequencies present respectively The spatial evolution of the system can be found by projecting the phase plot onto the spatial coordinate axis The maximum spatial excursion of each phase wrap corresponds to a sharply defined density maximum The density maxima occur near the turnaround points of the particle orbits and propagate slowly in radius to the outermost turning point set by the least bound particle The density maxima are therefore propagating density waves and are similar to the ring structures produced by Lynds and Toomre (1976) in their ring galaxy simulations A similar dynamical picture has been proposed for the evolution of large-scale structure in the universe (Zeldovich 1970 Doroshkevich et at 1980)

Figure 6 shows the radial velocityradius plane for the model presented in Figure 2 Here the phase wrapping nature of the shells can be clearly seen Note also that a much larger number of shells can be detected in this way than fromjust the particle distribution This is because the small number of particles used

is incapable of showing all the shells at a sufficiently high contrast to be seen

The phase wrapping interpretation of shell structures has a number of desirable features First the phase wraps are interleaved in radius as has been observed for the shell galaxies Second the range of the number of shells present around ellipticals is a simple consequence of the age of the event More shells will imply that a longer time has passed since the merger event given similar rates of shells production Shell producshytion rates can be estimated from the range in radii of the shells and hence the spread in periods present

The position in space of a particular shell can be calculated for a given potential by appreciating that the shells occur close to the radial turning points of the orbits Hence

Qdm t 2n(m - em) (2)

where Qd~ is the radial frequency ora particle with its radial turning point at radius dm dm is the distance of the shell from the center of the potential em is the orbital phase of particles with turning points at dm and m labels the shells by the number of orbital periods completed at time t by particles in the shell at that time The orbital phase of a particle is defined to be the time required to travel around its orbit to the nearest turning point in units of the orbital period hence -05s em s 05 If all the particles begin from rest then em = 0 for all particles and Qd

m(t2n) is an integer for all shells The variable m is the

proper shell number in that it is the total number of completed periods A second number n the observed number of shells is defined to be

n=m-t+1 (3)

where t is the total number of completed periods for particles in the outermost shell The outermost shell is therefore labeled

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

43 Photometry of Elliptical Galaxies 203

~ 095

I

SshyI

S 09

Figure 441 Galaxies with fine structure have bluer colors Here we plot the correlation between the fine-structure parameter ~ [equashytion (437)] and the color (B-V)-21

085 This is the galaxys color after corshyrection for the color-magnitude efshyfect (sect434) to absolute magnitude

08 0 2 4 6 8

MB = -21 [After Schweizer amp Seitzer (1992) from data kindly sup-

L plied by F Schweizer]

very hard to detect in late-type systems so we do not know how universal a phenomenon they are

A stellar system can display a sharp edge only if some parts of its phase space (see BT sect41) are very much more densely populated with stars than neighboring parts of phase space In the classical dynamical model of an elliptical (BT sect44) phase space is populated very smoothly Therefore the existence of ripples directly challenges the classical picture of ellipticals One likely possibility is that ellipticals acquire ripples late in life as a result of accreting material from a system within which there are relatively large grashydients in phase-space density Systems with large density gradients in phase space include disk galaxies and dwarf galaxies in a thin disk the phase-space density of stars peaks strongly around the locations of circular orbits while in a dwarf galaxy all stars move at approximately the systemic velocity so that there is only a small spread in velocity space

Numerical simulations suggest that ripples can indeed form when mateshyrial is accreted from either a disk galaxy or a dwarf system - see Barnes amp Hernquist (1992) for a review Moreover simulations have successfully reshyproduced the interleaved property of ripples described above Despite these successes significant uncertainties still surround the ripple phenomenon beshycause the available simulations have important limitations and it is not clear how probable their initial conditions are

Schweizer et al (1990) defined an index E that quantifies the amount of fine structure such as ripples that a galaxy possesses

E == S + 10g(1 + n) + J + B + X (437)

Here S measures the strength of the most prominent ripple on a scale of 0 to 3 n is the number of detected ripples J is the number of optical jets B is a measure of the boxiness of the galaxys isophotes on a scale of 0 to 4 X is o or 1 depending on whether the galaxys image shows an X structure For a sample of 69 nearby early-type galaxies E varies from 0 to 76 Notice that

VJe IJ- TIIEOIltpoundlJ

~r~d ampk~ 0 ~~1ud)~

0U- ccdd we M-la)W ~ wocdd J~ ~o~~

c- -$2 r

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 4: 3.4 cD Galaxies - Case Western Reserve University

lIow i7lt-you H1Ji- r(e ~aUshybull

JJoub ~ ~ e+~ ~ ~tL

Y O~ ueL~ so~ bull ~

~ ~ ~~~ ~rcAs middot

v~ middot ~~~ ~ 6if ~ ~fy~~ ~~~

bullbullbull

10 - w ~-y~ JaI~ ~ sMs

-s~ ~ ) ~ ~St7

4Ii7~~~sectefJf-f~ ~~~~~wd

J ~ of ~~~ ~ ~ ~ds ~ amp4 k ~ ~ ~ +

2J ~s~

PLATE 4

I

It

~

~

~ ~middot~~gti~H- 8~ ~~ ~~ _I- f

~~~ bull bull --00gt010004

bull ~

~IiitI~--~~~ ~ ~ ~

middoti ~~~ shy j ~bull

~ ~1 -

I I

FIG 4-NGC 3923 The top picture (Fig 4a) was made from thrce UK Schmidt I1Ia-J plates enhanced to show the NE shell Scale bar ~ is 10 The lower pictures (Figs 4h and 4() were mac from the same AAT plate printed through un~harp masks with diff~rent characteristics About l~ shells can be counted The dust cloud in Fig 4( is cal Scale bars are 5 and 2 in Fig 4h and 4c respecllvcI

M LI AND CARTER (see page 539)

06

_----_- -----shy

~

~ ~

~ ~ ~

bull

tl r-

MALIN AND CARTER (see page 538) ~

0 D

ltl

N

~ exgt

M2 6

10 14 18

~~

pound~

FIG I-Time evolution of the radial planar encounter between an exponential surface density disk and a 10 times more massive fixed and rigid isochrone potential The cross denotes the center of the isochrone potential and the disk is initially moving to the left The evolution is viewed from above the orbit plane (X -Y plane) and the times indicated are in units of the circular period at a radius of one scale length in the isochrone potential The bar under the initial time is 10 scale lengths long The disk as seen in the figure is rotating in a counterclockwise sense

2

10 14 18

FIG 2-Time evolution of the radia~ planar encounter of an exponential surface density disk and a 100 times more massive fixed and rigid isochrone potential Other figure parameters are as in Fig I

598

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

o D

ltl 602 rshyN

3 ~ co

- 0 ~ -1

-3

QUINN Vol 279

I I I

- ~

-

- ~

~

-~ii~~1~fllt-lt ~ - _ shy

- ~-~ bull

~

I I I I

o 5 10 15 20 radius

FIG 6-The radial velocity-radius plane for particles in the model shown iii Fig 2 at time 18 Radii are in units of the isochrone scale length and velocities are in units of (GM10a)1I2

dimensional system of test particles falling from rest into a fixed potential The extent of the system in the field introduces a spread of energies across the system For potentials other than the simple harmonic potential the range in energies corresshyponds to a range in periods with the most bound particles having the shortest periods Figure 5 shows the phase evolution of such a system As time passes the shorter period particles begin to lead the longer period ones and the system starts to wrap in the phase plane The wrapping proceeds at a rate determined by the range in periods present and the number of wraps at time t after the infall begins is simply

(1)

where Q is the radial frequency andQax aIld Qin are the maximum and minimum frequencies present respectively The spatial evolution of the system can be found by projecting the phase plot onto the spatial coordinate axis The maximum spatial excursion of each phase wrap corresponds to a sharply defined density maximum The density maxima occur near the turnaround points of the particle orbits and propagate slowly in radius to the outermost turning point set by the least bound particle The density maxima are therefore propagating density waves and are similar to the ring structures produced by Lynds and Toomre (1976) in their ring galaxy simulations A similar dynamical picture has been proposed for the evolution of large-scale structure in the universe (Zeldovich 1970 Doroshkevich et at 1980)

Figure 6 shows the radial velocityradius plane for the model presented in Figure 2 Here the phase wrapping nature of the shells can be clearly seen Note also that a much larger number of shells can be detected in this way than fromjust the particle distribution This is because the small number of particles used

is incapable of showing all the shells at a sufficiently high contrast to be seen

The phase wrapping interpretation of shell structures has a number of desirable features First the phase wraps are interleaved in radius as has been observed for the shell galaxies Second the range of the number of shells present around ellipticals is a simple consequence of the age of the event More shells will imply that a longer time has passed since the merger event given similar rates of shells production Shell producshytion rates can be estimated from the range in radii of the shells and hence the spread in periods present

The position in space of a particular shell can be calculated for a given potential by appreciating that the shells occur close to the radial turning points of the orbits Hence

Qdm t 2n(m - em) (2)

where Qd~ is the radial frequency ora particle with its radial turning point at radius dm dm is the distance of the shell from the center of the potential em is the orbital phase of particles with turning points at dm and m labels the shells by the number of orbital periods completed at time t by particles in the shell at that time The orbital phase of a particle is defined to be the time required to travel around its orbit to the nearest turning point in units of the orbital period hence -05s em s 05 If all the particles begin from rest then em = 0 for all particles and Qd

m(t2n) is an integer for all shells The variable m is the

proper shell number in that it is the total number of completed periods A second number n the observed number of shells is defined to be

n=m-t+1 (3)

where t is the total number of completed periods for particles in the outermost shell The outermost shell is therefore labeled

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

43 Photometry of Elliptical Galaxies 203

~ 095

I

SshyI

S 09

Figure 441 Galaxies with fine structure have bluer colors Here we plot the correlation between the fine-structure parameter ~ [equashytion (437)] and the color (B-V)-21

085 This is the galaxys color after corshyrection for the color-magnitude efshyfect (sect434) to absolute magnitude

08 0 2 4 6 8

MB = -21 [After Schweizer amp Seitzer (1992) from data kindly sup-

L plied by F Schweizer]

very hard to detect in late-type systems so we do not know how universal a phenomenon they are

A stellar system can display a sharp edge only if some parts of its phase space (see BT sect41) are very much more densely populated with stars than neighboring parts of phase space In the classical dynamical model of an elliptical (BT sect44) phase space is populated very smoothly Therefore the existence of ripples directly challenges the classical picture of ellipticals One likely possibility is that ellipticals acquire ripples late in life as a result of accreting material from a system within which there are relatively large grashydients in phase-space density Systems with large density gradients in phase space include disk galaxies and dwarf galaxies in a thin disk the phase-space density of stars peaks strongly around the locations of circular orbits while in a dwarf galaxy all stars move at approximately the systemic velocity so that there is only a small spread in velocity space

Numerical simulations suggest that ripples can indeed form when mateshyrial is accreted from either a disk galaxy or a dwarf system - see Barnes amp Hernquist (1992) for a review Moreover simulations have successfully reshyproduced the interleaved property of ripples described above Despite these successes significant uncertainties still surround the ripple phenomenon beshycause the available simulations have important limitations and it is not clear how probable their initial conditions are

Schweizer et al (1990) defined an index E that quantifies the amount of fine structure such as ripples that a galaxy possesses

E == S + 10g(1 + n) + J + B + X (437)

Here S measures the strength of the most prominent ripple on a scale of 0 to 3 n is the number of detected ripples J is the number of optical jets B is a measure of the boxiness of the galaxys isophotes on a scale of 0 to 4 X is o or 1 depending on whether the galaxys image shows an X structure For a sample of 69 nearby early-type galaxies E varies from 0 to 76 Notice that

VJe IJ- TIIEOIltpoundlJ

~r~d ampk~ 0 ~~1ud)~

0U- ccdd we M-la)W ~ wocdd J~ ~o~~

c- -$2 r

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 5: 3.4 cD Galaxies - Case Western Reserve University

10 - w ~-y~ JaI~ ~ sMs

-s~ ~ ) ~ ~St7

4Ii7~~~sectefJf-f~ ~~~~~wd

J ~ of ~~~ ~ ~ ~ds ~ amp4 k ~ ~ ~ +

2J ~s~

PLATE 4

I

It

~

~

~ ~middot~~gti~H- 8~ ~~ ~~ _I- f

~~~ bull bull --00gt010004

bull ~

~IiitI~--~~~ ~ ~ ~

middoti ~~~ shy j ~bull

~ ~1 -

I I

FIG 4-NGC 3923 The top picture (Fig 4a) was made from thrce UK Schmidt I1Ia-J plates enhanced to show the NE shell Scale bar ~ is 10 The lower pictures (Figs 4h and 4() were mac from the same AAT plate printed through un~harp masks with diff~rent characteristics About l~ shells can be counted The dust cloud in Fig 4( is cal Scale bars are 5 and 2 in Fig 4h and 4c respecllvcI

M LI AND CARTER (see page 539)

06

_----_- -----shy

~

~ ~

~ ~ ~

bull

tl r-

MALIN AND CARTER (see page 538) ~

0 D

ltl

N

~ exgt

M2 6

10 14 18

~~

pound~

FIG I-Time evolution of the radial planar encounter between an exponential surface density disk and a 10 times more massive fixed and rigid isochrone potential The cross denotes the center of the isochrone potential and the disk is initially moving to the left The evolution is viewed from above the orbit plane (X -Y plane) and the times indicated are in units of the circular period at a radius of one scale length in the isochrone potential The bar under the initial time is 10 scale lengths long The disk as seen in the figure is rotating in a counterclockwise sense

2

10 14 18

FIG 2-Time evolution of the radia~ planar encounter of an exponential surface density disk and a 100 times more massive fixed and rigid isochrone potential Other figure parameters are as in Fig I

598

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

o D

ltl 602 rshyN

3 ~ co

- 0 ~ -1

-3

QUINN Vol 279

I I I

- ~

-

- ~

~

-~ii~~1~fllt-lt ~ - _ shy

- ~-~ bull

~

I I I I

o 5 10 15 20 radius

FIG 6-The radial velocity-radius plane for particles in the model shown iii Fig 2 at time 18 Radii are in units of the isochrone scale length and velocities are in units of (GM10a)1I2

dimensional system of test particles falling from rest into a fixed potential The extent of the system in the field introduces a spread of energies across the system For potentials other than the simple harmonic potential the range in energies corresshyponds to a range in periods with the most bound particles having the shortest periods Figure 5 shows the phase evolution of such a system As time passes the shorter period particles begin to lead the longer period ones and the system starts to wrap in the phase plane The wrapping proceeds at a rate determined by the range in periods present and the number of wraps at time t after the infall begins is simply

(1)

where Q is the radial frequency andQax aIld Qin are the maximum and minimum frequencies present respectively The spatial evolution of the system can be found by projecting the phase plot onto the spatial coordinate axis The maximum spatial excursion of each phase wrap corresponds to a sharply defined density maximum The density maxima occur near the turnaround points of the particle orbits and propagate slowly in radius to the outermost turning point set by the least bound particle The density maxima are therefore propagating density waves and are similar to the ring structures produced by Lynds and Toomre (1976) in their ring galaxy simulations A similar dynamical picture has been proposed for the evolution of large-scale structure in the universe (Zeldovich 1970 Doroshkevich et at 1980)

Figure 6 shows the radial velocityradius plane for the model presented in Figure 2 Here the phase wrapping nature of the shells can be clearly seen Note also that a much larger number of shells can be detected in this way than fromjust the particle distribution This is because the small number of particles used

is incapable of showing all the shells at a sufficiently high contrast to be seen

The phase wrapping interpretation of shell structures has a number of desirable features First the phase wraps are interleaved in radius as has been observed for the shell galaxies Second the range of the number of shells present around ellipticals is a simple consequence of the age of the event More shells will imply that a longer time has passed since the merger event given similar rates of shells production Shell producshytion rates can be estimated from the range in radii of the shells and hence the spread in periods present

The position in space of a particular shell can be calculated for a given potential by appreciating that the shells occur close to the radial turning points of the orbits Hence

Qdm t 2n(m - em) (2)

where Qd~ is the radial frequency ora particle with its radial turning point at radius dm dm is the distance of the shell from the center of the potential em is the orbital phase of particles with turning points at dm and m labels the shells by the number of orbital periods completed at time t by particles in the shell at that time The orbital phase of a particle is defined to be the time required to travel around its orbit to the nearest turning point in units of the orbital period hence -05s em s 05 If all the particles begin from rest then em = 0 for all particles and Qd

m(t2n) is an integer for all shells The variable m is the

proper shell number in that it is the total number of completed periods A second number n the observed number of shells is defined to be

n=m-t+1 (3)

where t is the total number of completed periods for particles in the outermost shell The outermost shell is therefore labeled

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

43 Photometry of Elliptical Galaxies 203

~ 095

I

SshyI

S 09

Figure 441 Galaxies with fine structure have bluer colors Here we plot the correlation between the fine-structure parameter ~ [equashytion (437)] and the color (B-V)-21

085 This is the galaxys color after corshyrection for the color-magnitude efshyfect (sect434) to absolute magnitude

08 0 2 4 6 8

MB = -21 [After Schweizer amp Seitzer (1992) from data kindly sup-

L plied by F Schweizer]

very hard to detect in late-type systems so we do not know how universal a phenomenon they are

A stellar system can display a sharp edge only if some parts of its phase space (see BT sect41) are very much more densely populated with stars than neighboring parts of phase space In the classical dynamical model of an elliptical (BT sect44) phase space is populated very smoothly Therefore the existence of ripples directly challenges the classical picture of ellipticals One likely possibility is that ellipticals acquire ripples late in life as a result of accreting material from a system within which there are relatively large grashydients in phase-space density Systems with large density gradients in phase space include disk galaxies and dwarf galaxies in a thin disk the phase-space density of stars peaks strongly around the locations of circular orbits while in a dwarf galaxy all stars move at approximately the systemic velocity so that there is only a small spread in velocity space

Numerical simulations suggest that ripples can indeed form when mateshyrial is accreted from either a disk galaxy or a dwarf system - see Barnes amp Hernquist (1992) for a review Moreover simulations have successfully reshyproduced the interleaved property of ripples described above Despite these successes significant uncertainties still surround the ripple phenomenon beshycause the available simulations have important limitations and it is not clear how probable their initial conditions are

Schweizer et al (1990) defined an index E that quantifies the amount of fine structure such as ripples that a galaxy possesses

E == S + 10g(1 + n) + J + B + X (437)

Here S measures the strength of the most prominent ripple on a scale of 0 to 3 n is the number of detected ripples J is the number of optical jets B is a measure of the boxiness of the galaxys isophotes on a scale of 0 to 4 X is o or 1 depending on whether the galaxys image shows an X structure For a sample of 69 nearby early-type galaxies E varies from 0 to 76 Notice that

VJe IJ- TIIEOIltpoundlJ

~r~d ampk~ 0 ~~1ud)~

0U- ccdd we M-la)W ~ wocdd J~ ~o~~

c- -$2 r

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 6: 3.4 cD Galaxies - Case Western Reserve University

PLATE 4

I

It

~

~

~ ~middot~~gti~H- 8~ ~~ ~~ _I- f

~~~ bull bull --00gt010004

bull ~

~IiitI~--~~~ ~ ~ ~

middoti ~~~ shy j ~bull

~ ~1 -

I I

FIG 4-NGC 3923 The top picture (Fig 4a) was made from thrce UK Schmidt I1Ia-J plates enhanced to show the NE shell Scale bar ~ is 10 The lower pictures (Figs 4h and 4() were mac from the same AAT plate printed through un~harp masks with diff~rent characteristics About l~ shells can be counted The dust cloud in Fig 4( is cal Scale bars are 5 and 2 in Fig 4h and 4c respecllvcI

M LI AND CARTER (see page 539)

06

_----_- -----shy

~

~ ~

~ ~ ~

bull

tl r-

MALIN AND CARTER (see page 538) ~

0 D

ltl

N

~ exgt

M2 6

10 14 18

~~

pound~

FIG I-Time evolution of the radial planar encounter between an exponential surface density disk and a 10 times more massive fixed and rigid isochrone potential The cross denotes the center of the isochrone potential and the disk is initially moving to the left The evolution is viewed from above the orbit plane (X -Y plane) and the times indicated are in units of the circular period at a radius of one scale length in the isochrone potential The bar under the initial time is 10 scale lengths long The disk as seen in the figure is rotating in a counterclockwise sense

2

10 14 18

FIG 2-Time evolution of the radia~ planar encounter of an exponential surface density disk and a 100 times more massive fixed and rigid isochrone potential Other figure parameters are as in Fig I

598

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

o D

ltl 602 rshyN

3 ~ co

- 0 ~ -1

-3

QUINN Vol 279

I I I

- ~

-

- ~

~

-~ii~~1~fllt-lt ~ - _ shy

- ~-~ bull

~

I I I I

o 5 10 15 20 radius

FIG 6-The radial velocity-radius plane for particles in the model shown iii Fig 2 at time 18 Radii are in units of the isochrone scale length and velocities are in units of (GM10a)1I2

dimensional system of test particles falling from rest into a fixed potential The extent of the system in the field introduces a spread of energies across the system For potentials other than the simple harmonic potential the range in energies corresshyponds to a range in periods with the most bound particles having the shortest periods Figure 5 shows the phase evolution of such a system As time passes the shorter period particles begin to lead the longer period ones and the system starts to wrap in the phase plane The wrapping proceeds at a rate determined by the range in periods present and the number of wraps at time t after the infall begins is simply

(1)

where Q is the radial frequency andQax aIld Qin are the maximum and minimum frequencies present respectively The spatial evolution of the system can be found by projecting the phase plot onto the spatial coordinate axis The maximum spatial excursion of each phase wrap corresponds to a sharply defined density maximum The density maxima occur near the turnaround points of the particle orbits and propagate slowly in radius to the outermost turning point set by the least bound particle The density maxima are therefore propagating density waves and are similar to the ring structures produced by Lynds and Toomre (1976) in their ring galaxy simulations A similar dynamical picture has been proposed for the evolution of large-scale structure in the universe (Zeldovich 1970 Doroshkevich et at 1980)

Figure 6 shows the radial velocityradius plane for the model presented in Figure 2 Here the phase wrapping nature of the shells can be clearly seen Note also that a much larger number of shells can be detected in this way than fromjust the particle distribution This is because the small number of particles used

is incapable of showing all the shells at a sufficiently high contrast to be seen

The phase wrapping interpretation of shell structures has a number of desirable features First the phase wraps are interleaved in radius as has been observed for the shell galaxies Second the range of the number of shells present around ellipticals is a simple consequence of the age of the event More shells will imply that a longer time has passed since the merger event given similar rates of shells production Shell producshytion rates can be estimated from the range in radii of the shells and hence the spread in periods present

The position in space of a particular shell can be calculated for a given potential by appreciating that the shells occur close to the radial turning points of the orbits Hence

Qdm t 2n(m - em) (2)

where Qd~ is the radial frequency ora particle with its radial turning point at radius dm dm is the distance of the shell from the center of the potential em is the orbital phase of particles with turning points at dm and m labels the shells by the number of orbital periods completed at time t by particles in the shell at that time The orbital phase of a particle is defined to be the time required to travel around its orbit to the nearest turning point in units of the orbital period hence -05s em s 05 If all the particles begin from rest then em = 0 for all particles and Qd

m(t2n) is an integer for all shells The variable m is the

proper shell number in that it is the total number of completed periods A second number n the observed number of shells is defined to be

n=m-t+1 (3)

where t is the total number of completed periods for particles in the outermost shell The outermost shell is therefore labeled

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

43 Photometry of Elliptical Galaxies 203

~ 095

I

SshyI

S 09

Figure 441 Galaxies with fine structure have bluer colors Here we plot the correlation between the fine-structure parameter ~ [equashytion (437)] and the color (B-V)-21

085 This is the galaxys color after corshyrection for the color-magnitude efshyfect (sect434) to absolute magnitude

08 0 2 4 6 8

MB = -21 [After Schweizer amp Seitzer (1992) from data kindly sup-

L plied by F Schweizer]

very hard to detect in late-type systems so we do not know how universal a phenomenon they are

A stellar system can display a sharp edge only if some parts of its phase space (see BT sect41) are very much more densely populated with stars than neighboring parts of phase space In the classical dynamical model of an elliptical (BT sect44) phase space is populated very smoothly Therefore the existence of ripples directly challenges the classical picture of ellipticals One likely possibility is that ellipticals acquire ripples late in life as a result of accreting material from a system within which there are relatively large grashydients in phase-space density Systems with large density gradients in phase space include disk galaxies and dwarf galaxies in a thin disk the phase-space density of stars peaks strongly around the locations of circular orbits while in a dwarf galaxy all stars move at approximately the systemic velocity so that there is only a small spread in velocity space

Numerical simulations suggest that ripples can indeed form when mateshyrial is accreted from either a disk galaxy or a dwarf system - see Barnes amp Hernquist (1992) for a review Moreover simulations have successfully reshyproduced the interleaved property of ripples described above Despite these successes significant uncertainties still surround the ripple phenomenon beshycause the available simulations have important limitations and it is not clear how probable their initial conditions are

Schweizer et al (1990) defined an index E that quantifies the amount of fine structure such as ripples that a galaxy possesses

E == S + 10g(1 + n) + J + B + X (437)

Here S measures the strength of the most prominent ripple on a scale of 0 to 3 n is the number of detected ripples J is the number of optical jets B is a measure of the boxiness of the galaxys isophotes on a scale of 0 to 4 X is o or 1 depending on whether the galaxys image shows an X structure For a sample of 69 nearby early-type galaxies E varies from 0 to 76 Notice that

VJe IJ- TIIEOIltpoundlJ

~r~d ampk~ 0 ~~1ud)~

0U- ccdd we M-la)W ~ wocdd J~ ~o~~

c- -$2 r

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 7: 3.4 cD Galaxies - Case Western Reserve University

06

_----_- -----shy

~

~ ~

~ ~ ~

bull

tl r-

MALIN AND CARTER (see page 538) ~

0 D

ltl

N

~ exgt

M2 6

10 14 18

~~

pound~

FIG I-Time evolution of the radial planar encounter between an exponential surface density disk and a 10 times more massive fixed and rigid isochrone potential The cross denotes the center of the isochrone potential and the disk is initially moving to the left The evolution is viewed from above the orbit plane (X -Y plane) and the times indicated are in units of the circular period at a radius of one scale length in the isochrone potential The bar under the initial time is 10 scale lengths long The disk as seen in the figure is rotating in a counterclockwise sense

2

10 14 18

FIG 2-Time evolution of the radia~ planar encounter of an exponential surface density disk and a 100 times more massive fixed and rigid isochrone potential Other figure parameters are as in Fig I

598

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

o D

ltl 602 rshyN

3 ~ co

- 0 ~ -1

-3

QUINN Vol 279

I I I

- ~

-

- ~

~

-~ii~~1~fllt-lt ~ - _ shy

- ~-~ bull

~

I I I I

o 5 10 15 20 radius

FIG 6-The radial velocity-radius plane for particles in the model shown iii Fig 2 at time 18 Radii are in units of the isochrone scale length and velocities are in units of (GM10a)1I2

dimensional system of test particles falling from rest into a fixed potential The extent of the system in the field introduces a spread of energies across the system For potentials other than the simple harmonic potential the range in energies corresshyponds to a range in periods with the most bound particles having the shortest periods Figure 5 shows the phase evolution of such a system As time passes the shorter period particles begin to lead the longer period ones and the system starts to wrap in the phase plane The wrapping proceeds at a rate determined by the range in periods present and the number of wraps at time t after the infall begins is simply

(1)

where Q is the radial frequency andQax aIld Qin are the maximum and minimum frequencies present respectively The spatial evolution of the system can be found by projecting the phase plot onto the spatial coordinate axis The maximum spatial excursion of each phase wrap corresponds to a sharply defined density maximum The density maxima occur near the turnaround points of the particle orbits and propagate slowly in radius to the outermost turning point set by the least bound particle The density maxima are therefore propagating density waves and are similar to the ring structures produced by Lynds and Toomre (1976) in their ring galaxy simulations A similar dynamical picture has been proposed for the evolution of large-scale structure in the universe (Zeldovich 1970 Doroshkevich et at 1980)

Figure 6 shows the radial velocityradius plane for the model presented in Figure 2 Here the phase wrapping nature of the shells can be clearly seen Note also that a much larger number of shells can be detected in this way than fromjust the particle distribution This is because the small number of particles used

is incapable of showing all the shells at a sufficiently high contrast to be seen

The phase wrapping interpretation of shell structures has a number of desirable features First the phase wraps are interleaved in radius as has been observed for the shell galaxies Second the range of the number of shells present around ellipticals is a simple consequence of the age of the event More shells will imply that a longer time has passed since the merger event given similar rates of shells production Shell producshytion rates can be estimated from the range in radii of the shells and hence the spread in periods present

The position in space of a particular shell can be calculated for a given potential by appreciating that the shells occur close to the radial turning points of the orbits Hence

Qdm t 2n(m - em) (2)

where Qd~ is the radial frequency ora particle with its radial turning point at radius dm dm is the distance of the shell from the center of the potential em is the orbital phase of particles with turning points at dm and m labels the shells by the number of orbital periods completed at time t by particles in the shell at that time The orbital phase of a particle is defined to be the time required to travel around its orbit to the nearest turning point in units of the orbital period hence -05s em s 05 If all the particles begin from rest then em = 0 for all particles and Qd

m(t2n) is an integer for all shells The variable m is the

proper shell number in that it is the total number of completed periods A second number n the observed number of shells is defined to be

n=m-t+1 (3)

where t is the total number of completed periods for particles in the outermost shell The outermost shell is therefore labeled

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

43 Photometry of Elliptical Galaxies 203

~ 095

I

SshyI

S 09

Figure 441 Galaxies with fine structure have bluer colors Here we plot the correlation between the fine-structure parameter ~ [equashytion (437)] and the color (B-V)-21

085 This is the galaxys color after corshyrection for the color-magnitude efshyfect (sect434) to absolute magnitude

08 0 2 4 6 8

MB = -21 [After Schweizer amp Seitzer (1992) from data kindly sup-

L plied by F Schweizer]

very hard to detect in late-type systems so we do not know how universal a phenomenon they are

A stellar system can display a sharp edge only if some parts of its phase space (see BT sect41) are very much more densely populated with stars than neighboring parts of phase space In the classical dynamical model of an elliptical (BT sect44) phase space is populated very smoothly Therefore the existence of ripples directly challenges the classical picture of ellipticals One likely possibility is that ellipticals acquire ripples late in life as a result of accreting material from a system within which there are relatively large grashydients in phase-space density Systems with large density gradients in phase space include disk galaxies and dwarf galaxies in a thin disk the phase-space density of stars peaks strongly around the locations of circular orbits while in a dwarf galaxy all stars move at approximately the systemic velocity so that there is only a small spread in velocity space

Numerical simulations suggest that ripples can indeed form when mateshyrial is accreted from either a disk galaxy or a dwarf system - see Barnes amp Hernquist (1992) for a review Moreover simulations have successfully reshyproduced the interleaved property of ripples described above Despite these successes significant uncertainties still surround the ripple phenomenon beshycause the available simulations have important limitations and it is not clear how probable their initial conditions are

Schweizer et al (1990) defined an index E that quantifies the amount of fine structure such as ripples that a galaxy possesses

E == S + 10g(1 + n) + J + B + X (437)

Here S measures the strength of the most prominent ripple on a scale of 0 to 3 n is the number of detected ripples J is the number of optical jets B is a measure of the boxiness of the galaxys isophotes on a scale of 0 to 4 X is o or 1 depending on whether the galaxys image shows an X structure For a sample of 69 nearby early-type galaxies E varies from 0 to 76 Notice that

VJe IJ- TIIEOIltpoundlJ

~r~d ampk~ 0 ~~1ud)~

0U- ccdd we M-la)W ~ wocdd J~ ~o~~

c- -$2 r

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 8: 3.4 cD Galaxies - Case Western Reserve University

0 D

ltl

N

~ exgt

M2 6

10 14 18

~~

pound~

FIG I-Time evolution of the radial planar encounter between an exponential surface density disk and a 10 times more massive fixed and rigid isochrone potential The cross denotes the center of the isochrone potential and the disk is initially moving to the left The evolution is viewed from above the orbit plane (X -Y plane) and the times indicated are in units of the circular period at a radius of one scale length in the isochrone potential The bar under the initial time is 10 scale lengths long The disk as seen in the figure is rotating in a counterclockwise sense

2

10 14 18

FIG 2-Time evolution of the radia~ planar encounter of an exponential surface density disk and a 100 times more massive fixed and rigid isochrone potential Other figure parameters are as in Fig I

598

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

o D

ltl 602 rshyN

3 ~ co

- 0 ~ -1

-3

QUINN Vol 279

I I I

- ~

-

- ~

~

-~ii~~1~fllt-lt ~ - _ shy

- ~-~ bull

~

I I I I

o 5 10 15 20 radius

FIG 6-The radial velocity-radius plane for particles in the model shown iii Fig 2 at time 18 Radii are in units of the isochrone scale length and velocities are in units of (GM10a)1I2

dimensional system of test particles falling from rest into a fixed potential The extent of the system in the field introduces a spread of energies across the system For potentials other than the simple harmonic potential the range in energies corresshyponds to a range in periods with the most bound particles having the shortest periods Figure 5 shows the phase evolution of such a system As time passes the shorter period particles begin to lead the longer period ones and the system starts to wrap in the phase plane The wrapping proceeds at a rate determined by the range in periods present and the number of wraps at time t after the infall begins is simply

(1)

where Q is the radial frequency andQax aIld Qin are the maximum and minimum frequencies present respectively The spatial evolution of the system can be found by projecting the phase plot onto the spatial coordinate axis The maximum spatial excursion of each phase wrap corresponds to a sharply defined density maximum The density maxima occur near the turnaround points of the particle orbits and propagate slowly in radius to the outermost turning point set by the least bound particle The density maxima are therefore propagating density waves and are similar to the ring structures produced by Lynds and Toomre (1976) in their ring galaxy simulations A similar dynamical picture has been proposed for the evolution of large-scale structure in the universe (Zeldovich 1970 Doroshkevich et at 1980)

Figure 6 shows the radial velocityradius plane for the model presented in Figure 2 Here the phase wrapping nature of the shells can be clearly seen Note also that a much larger number of shells can be detected in this way than fromjust the particle distribution This is because the small number of particles used

is incapable of showing all the shells at a sufficiently high contrast to be seen

The phase wrapping interpretation of shell structures has a number of desirable features First the phase wraps are interleaved in radius as has been observed for the shell galaxies Second the range of the number of shells present around ellipticals is a simple consequence of the age of the event More shells will imply that a longer time has passed since the merger event given similar rates of shells production Shell producshytion rates can be estimated from the range in radii of the shells and hence the spread in periods present

The position in space of a particular shell can be calculated for a given potential by appreciating that the shells occur close to the radial turning points of the orbits Hence

Qdm t 2n(m - em) (2)

where Qd~ is the radial frequency ora particle with its radial turning point at radius dm dm is the distance of the shell from the center of the potential em is the orbital phase of particles with turning points at dm and m labels the shells by the number of orbital periods completed at time t by particles in the shell at that time The orbital phase of a particle is defined to be the time required to travel around its orbit to the nearest turning point in units of the orbital period hence -05s em s 05 If all the particles begin from rest then em = 0 for all particles and Qd

m(t2n) is an integer for all shells The variable m is the

proper shell number in that it is the total number of completed periods A second number n the observed number of shells is defined to be

n=m-t+1 (3)

where t is the total number of completed periods for particles in the outermost shell The outermost shell is therefore labeled

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

43 Photometry of Elliptical Galaxies 203

~ 095

I

SshyI

S 09

Figure 441 Galaxies with fine structure have bluer colors Here we plot the correlation between the fine-structure parameter ~ [equashytion (437)] and the color (B-V)-21

085 This is the galaxys color after corshyrection for the color-magnitude efshyfect (sect434) to absolute magnitude

08 0 2 4 6 8

MB = -21 [After Schweizer amp Seitzer (1992) from data kindly sup-

L plied by F Schweizer]

very hard to detect in late-type systems so we do not know how universal a phenomenon they are

A stellar system can display a sharp edge only if some parts of its phase space (see BT sect41) are very much more densely populated with stars than neighboring parts of phase space In the classical dynamical model of an elliptical (BT sect44) phase space is populated very smoothly Therefore the existence of ripples directly challenges the classical picture of ellipticals One likely possibility is that ellipticals acquire ripples late in life as a result of accreting material from a system within which there are relatively large grashydients in phase-space density Systems with large density gradients in phase space include disk galaxies and dwarf galaxies in a thin disk the phase-space density of stars peaks strongly around the locations of circular orbits while in a dwarf galaxy all stars move at approximately the systemic velocity so that there is only a small spread in velocity space

Numerical simulations suggest that ripples can indeed form when mateshyrial is accreted from either a disk galaxy or a dwarf system - see Barnes amp Hernquist (1992) for a review Moreover simulations have successfully reshyproduced the interleaved property of ripples described above Despite these successes significant uncertainties still surround the ripple phenomenon beshycause the available simulations have important limitations and it is not clear how probable their initial conditions are

Schweizer et al (1990) defined an index E that quantifies the amount of fine structure such as ripples that a galaxy possesses

E == S + 10g(1 + n) + J + B + X (437)

Here S measures the strength of the most prominent ripple on a scale of 0 to 3 n is the number of detected ripples J is the number of optical jets B is a measure of the boxiness of the galaxys isophotes on a scale of 0 to 4 X is o or 1 depending on whether the galaxys image shows an X structure For a sample of 69 nearby early-type galaxies E varies from 0 to 76 Notice that

VJe IJ- TIIEOIltpoundlJ

~r~d ampk~ 0 ~~1ud)~

0U- ccdd we M-la)W ~ wocdd J~ ~o~~

c- -$2 r

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 9: 3.4 cD Galaxies - Case Western Reserve University

o D

ltl 602 rshyN

3 ~ co

- 0 ~ -1

-3

QUINN Vol 279

I I I

- ~

-

- ~

~

-~ii~~1~fllt-lt ~ - _ shy

- ~-~ bull

~

I I I I

o 5 10 15 20 radius

FIG 6-The radial velocity-radius plane for particles in the model shown iii Fig 2 at time 18 Radii are in units of the isochrone scale length and velocities are in units of (GM10a)1I2

dimensional system of test particles falling from rest into a fixed potential The extent of the system in the field introduces a spread of energies across the system For potentials other than the simple harmonic potential the range in energies corresshyponds to a range in periods with the most bound particles having the shortest periods Figure 5 shows the phase evolution of such a system As time passes the shorter period particles begin to lead the longer period ones and the system starts to wrap in the phase plane The wrapping proceeds at a rate determined by the range in periods present and the number of wraps at time t after the infall begins is simply

(1)

where Q is the radial frequency andQax aIld Qin are the maximum and minimum frequencies present respectively The spatial evolution of the system can be found by projecting the phase plot onto the spatial coordinate axis The maximum spatial excursion of each phase wrap corresponds to a sharply defined density maximum The density maxima occur near the turnaround points of the particle orbits and propagate slowly in radius to the outermost turning point set by the least bound particle The density maxima are therefore propagating density waves and are similar to the ring structures produced by Lynds and Toomre (1976) in their ring galaxy simulations A similar dynamical picture has been proposed for the evolution of large-scale structure in the universe (Zeldovich 1970 Doroshkevich et at 1980)

Figure 6 shows the radial velocityradius plane for the model presented in Figure 2 Here the phase wrapping nature of the shells can be clearly seen Note also that a much larger number of shells can be detected in this way than fromjust the particle distribution This is because the small number of particles used

is incapable of showing all the shells at a sufficiently high contrast to be seen

The phase wrapping interpretation of shell structures has a number of desirable features First the phase wraps are interleaved in radius as has been observed for the shell galaxies Second the range of the number of shells present around ellipticals is a simple consequence of the age of the event More shells will imply that a longer time has passed since the merger event given similar rates of shells production Shell producshytion rates can be estimated from the range in radii of the shells and hence the spread in periods present

The position in space of a particular shell can be calculated for a given potential by appreciating that the shells occur close to the radial turning points of the orbits Hence

Qdm t 2n(m - em) (2)

where Qd~ is the radial frequency ora particle with its radial turning point at radius dm dm is the distance of the shell from the center of the potential em is the orbital phase of particles with turning points at dm and m labels the shells by the number of orbital periods completed at time t by particles in the shell at that time The orbital phase of a particle is defined to be the time required to travel around its orbit to the nearest turning point in units of the orbital period hence -05s em s 05 If all the particles begin from rest then em = 0 for all particles and Qd

m(t2n) is an integer for all shells The variable m is the

proper shell number in that it is the total number of completed periods A second number n the observed number of shells is defined to be

n=m-t+1 (3)

where t is the total number of completed periods for particles in the outermost shell The outermost shell is therefore labeled

copy American Astronomical Society bull Provided by the NASA Astrophysics Data System

43 Photometry of Elliptical Galaxies 203

~ 095

I

SshyI

S 09

Figure 441 Galaxies with fine structure have bluer colors Here we plot the correlation between the fine-structure parameter ~ [equashytion (437)] and the color (B-V)-21

085 This is the galaxys color after corshyrection for the color-magnitude efshyfect (sect434) to absolute magnitude

08 0 2 4 6 8

MB = -21 [After Schweizer amp Seitzer (1992) from data kindly sup-

L plied by F Schweizer]

very hard to detect in late-type systems so we do not know how universal a phenomenon they are

A stellar system can display a sharp edge only if some parts of its phase space (see BT sect41) are very much more densely populated with stars than neighboring parts of phase space In the classical dynamical model of an elliptical (BT sect44) phase space is populated very smoothly Therefore the existence of ripples directly challenges the classical picture of ellipticals One likely possibility is that ellipticals acquire ripples late in life as a result of accreting material from a system within which there are relatively large grashydients in phase-space density Systems with large density gradients in phase space include disk galaxies and dwarf galaxies in a thin disk the phase-space density of stars peaks strongly around the locations of circular orbits while in a dwarf galaxy all stars move at approximately the systemic velocity so that there is only a small spread in velocity space

Numerical simulations suggest that ripples can indeed form when mateshyrial is accreted from either a disk galaxy or a dwarf system - see Barnes amp Hernquist (1992) for a review Moreover simulations have successfully reshyproduced the interleaved property of ripples described above Despite these successes significant uncertainties still surround the ripple phenomenon beshycause the available simulations have important limitations and it is not clear how probable their initial conditions are

Schweizer et al (1990) defined an index E that quantifies the amount of fine structure such as ripples that a galaxy possesses

E == S + 10g(1 + n) + J + B + X (437)

Here S measures the strength of the most prominent ripple on a scale of 0 to 3 n is the number of detected ripples J is the number of optical jets B is a measure of the boxiness of the galaxys isophotes on a scale of 0 to 4 X is o or 1 depending on whether the galaxys image shows an X structure For a sample of 69 nearby early-type galaxies E varies from 0 to 76 Notice that

VJe IJ- TIIEOIltpoundlJ

~r~d ampk~ 0 ~~1ud)~

0U- ccdd we M-la)W ~ wocdd J~ ~o~~

c- -$2 r

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 10: 3.4 cD Galaxies - Case Western Reserve University

43 Photometry of Elliptical Galaxies 203

~ 095

I

SshyI

S 09

Figure 441 Galaxies with fine structure have bluer colors Here we plot the correlation between the fine-structure parameter ~ [equashytion (437)] and the color (B-V)-21

085 This is the galaxys color after corshyrection for the color-magnitude efshyfect (sect434) to absolute magnitude

08 0 2 4 6 8

MB = -21 [After Schweizer amp Seitzer (1992) from data kindly sup-

L plied by F Schweizer]

very hard to detect in late-type systems so we do not know how universal a phenomenon they are

A stellar system can display a sharp edge only if some parts of its phase space (see BT sect41) are very much more densely populated with stars than neighboring parts of phase space In the classical dynamical model of an elliptical (BT sect44) phase space is populated very smoothly Therefore the existence of ripples directly challenges the classical picture of ellipticals One likely possibility is that ellipticals acquire ripples late in life as a result of accreting material from a system within which there are relatively large grashydients in phase-space density Systems with large density gradients in phase space include disk galaxies and dwarf galaxies in a thin disk the phase-space density of stars peaks strongly around the locations of circular orbits while in a dwarf galaxy all stars move at approximately the systemic velocity so that there is only a small spread in velocity space

Numerical simulations suggest that ripples can indeed form when mateshyrial is accreted from either a disk galaxy or a dwarf system - see Barnes amp Hernquist (1992) for a review Moreover simulations have successfully reshyproduced the interleaved property of ripples described above Despite these successes significant uncertainties still surround the ripple phenomenon beshycause the available simulations have important limitations and it is not clear how probable their initial conditions are

Schweizer et al (1990) defined an index E that quantifies the amount of fine structure such as ripples that a galaxy possesses

E == S + 10g(1 + n) + J + B + X (437)

Here S measures the strength of the most prominent ripple on a scale of 0 to 3 n is the number of detected ripples J is the number of optical jets B is a measure of the boxiness of the galaxys isophotes on a scale of 0 to 4 X is o or 1 depending on whether the galaxys image shows an X structure For a sample of 69 nearby early-type galaxies E varies from 0 to 76 Notice that

VJe IJ- TIIEOIltpoundlJ

~r~d ampk~ 0 ~~1ud)~

0U- ccdd we M-la)W ~ wocdd J~ ~o~~

c- -$2 r

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 11: 3.4 cD Galaxies - Case Western Reserve University

VJe IJ- TIIEOIltpoundlJ

~r~d ampk~ 0 ~~1ud)~

0U- ccdd we M-la)W ~ wocdd J~ ~o~~

c- -$2 r

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 12: 3.4 cD Galaxies - Case Western Reserve University

bull bull

bull

Stk~ 0 ~ - hJs -

~cJ~amp~wdX ~ amp ~ ~v amputJ

~

(~I (~Jt0(

~) ~~~J~ Q-~01

tk S7A 0 ~ toLoy (~s t4 ~$

~~T~~)

CClA ~ hLCVICU4 Z4JA-~ ~

+ blfALAvn ~ ~ ~ ~

~ ~ daJ-euy ~~ bull

~~C2

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 13: 3.4 cD Galaxies - Case Western Reserve University

I 15

10 r- u

eshy 05-

l

0-00 000 l

-05 v

-8 ~10 -12 -14 -16 -18 -20 -22 -24

j1MB

Figure 2329 Normal ellipticals and dwarf systems define separate seshyquences of effective radius as a function of MB The bulges of spiral galaxshyies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399462 1992)

c~ ~ s~0 dE D- dSrA

~)

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 14: 3.4 cD Galaxies - Case Western Reserve University

17 18 19

ClI 20 U (1)

21 ~ 22 ~ S 23 I

Itl - 24

J

25 26 27 28

~

bull gEs () lnt Es iil eEs

Bright dEs bull v dSphs 0 Bulges

normal Es ~

dwarfEs

v v

7

VI

IlII 0

1iJ

-6 -8 -10 -12 -14 -16 -18 -20 -22 -24 MB

Figure 2330 Normal ellipticals and dwarf systems define separate seshyquences of average surface brightness versus M B The bulges of spiral galaxies lie along the sequence of normal ellipticals (Data from Bender et al Ap J 399 462 1992)

B~ dfyhaL5 Moe- LOUJC7lt (~)

~s~~~

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 15: 3.4 cD Galaxies - Case Western Reserve University

7JoJe ~ for 4~

2)~ fe ~amp ~ pound ~ ~ = cuJ-~ veLo~ c40~o~

J~-vadsc ~~ ~ kcvugtsamp 8- ~~ (0 ~ _~ xL~~ r- ~ )

J(e 0 a 9shy

UJ~ ~ ~ ~daLW~ ~

e) ~eL t~~ ~ Re ~~ gt) ) L e

~~

~ ~~~~ ev- (~) e+z-~

caA~ ~~j~

ie y Ae (3middot36 ~ a~ t

~c 8kd ~J

~~~71

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 16: 3.4 cD Galaxies - Case Western Reserve University

Chapter 4 Morphology of Galaxies206

)-I middotmiddot1middot IrtJ

S I bull bull ~~ ~~ ~ I

~ ~

0 0 b 0

0 co

18 0 ~ ~ i 0 bullbull J~ ~ -r

OJ 20 i ~~middoti ~ shy-lt v

22

01 1 10 -16 -18 -20 -22 -24 -26 Rkpc Le

Figure 443 Correlations between four shape-independent parameters of elliptical galaxshyies The parameters are the effective radius Re the mean surface brightness within Re (I)e the central velocity dispersion 00 and Le the luminosity in Djorgovskis G band interior to Re The luminosity and the surface brightness are expressed in magnitudes and in magnitudes per square arcsecond respectively [From data published in Djorgovski amp Davis (1987)J

The lower right panel of Figure 443 shows the correlation of which this is the mean regression

More luminous ellipticals have larger central velocity dispersions The upper right panel in Figure 443 illustrates this correlation which is called the Faber-Jackson relation after its discoverers (Faber amp Jackson 1976) Quantitatively one has

(438)

Since 00 is correlated with Le and Le is correlated with Re it follows that 00

must be correlated with R e The top left panel in Figure 443 displays this correlation

Since 00 is strongly correlated with line-strengths and colors the exisshytence of a host of additional correlations involving Mgz B - V etc is implied by the top two panels of Figure 443 One of the earliest of these to be discovshyered was the color-magnitude relation Faber (1973) showed that more luminous elliptical galaxies have stronger absorption lines and Visvanathan amp Sandage (1977) showed that more luminous galaxies are redder

In all the correlations of Figure 443 there is cosmic scatter the scatshyter of the points about the mean relations is larger than can be accounted

43 Photometry of Ellipti

Box 42 PI

Given M points x Cltraquo in d-dimensional unit vecto nearly as possible on theuro respect to p and the con

subject to the constrair multipliers this problem

0= LM

[(n ltgt=1

M

0= 2)n ltgt=1

where gt is the undeterr using the last equation t Then we find that

0= L(n ltgt

This equation can be rel

Thus the required vector A One may show (se( values taken by S for n the desired n is the eige

In order to minimiz have been scaled such th

for by measurement erron als between the positions correlated For example 1

relation in the top right pa magnitude relation in the of both the basic correlat is useful to imagine each

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -

Page 17: 3.4 cD Galaxies - Case Western Reserve University

Chapter 4 Morphology of Galaxies208

9

~ bD 0

+ 1 1

8 ~ ~

~ ~(~~

-

V

D co 0

7~ - Figure 444 An edge-on view of the fundamental plane as defined by the data of Figure 443 Note how much narrower is the distribution of points

6-1 a 2 than in either of the left-hand panels

Rlkpc of Figure 443

dimensional space The Cartesian coordinates Xi of a point in this space are the three numbers of the set (logRe (1)elogOO) where (1)e is in units of JLB 14 If there were no correlations between the variables the points of individual galaxies would be fairly uniformly distributed within a cuboidal region of our three-dimensional space Correlations between the variables will confine the data points to a sub-volume of this cuboid For example if two of the variables are genuinely independent and the third dependent on these two the points will be confined to a plane while if there is only one independent variable the points will lie on a line Our first step in analyzing such data is to ask whether the points are nearly confined to a plane if we can find a suitable plane we can further enquire whether the points lie on a line within that plane In statistics books this type of investigation is called principal component analysis Box 42 explains what has to be done

For the data in Figure 443 one finds that the points are as nearly conshyfined to a plane as observational errors allow in the absence of observational errors they might lie precisely on a plane If the position vector of a galaxy in our three-space is g = (log Re (I)e log 00) where R e is measured in kpc (I)e in units of JLB and 00 in km s-l then the equation of this fundamental plane is nmiddot g = 1 where n = (-065022086)

Naturally one wants to see the fundamental plane One way to do this is to choose an axis for example the Re-axis as the horizontal axis and rotate the space about this axis until the plane appears edge-on In this orientation the planes normal n lies in the plane of the projection which is spanned by the unit vector eR that runs parallel to Re-axis and some other orthogonal unit vector e Thus n = oeR + 3e where 0 and 3 are suitable numbers Comparing this with the value of n given above we see that 3e = 022e[ + 086e Thus the fundamental plane will be seen edge-on if we plot log Re against 022( (1)e JLB) + 086 log 00 Actually it is conventional to plot 026( (1)e JLB) + log 00 which is simply 1086 times the linear combination we have derived Figure 444 shows this plot The

14 Le need not be included in the set since it is related to Re and (I)e by Le = 1r(I)eR where L is measured in W Ie in Wm-2sterad-1 and Re in m

43 Photometry of Ellipti

data points nearly lie on fundamental plane The eo

logR

is also an equation for thE plying nmiddot g = 1 by 1065

The D n - 00 correlatio measured photometric pa virtue of the fundamental which the mean surface bl terms of a fiducial surfacE surface brightness than fo luminosity To quantify tl same surface-brightness pI 1(R) = 1ef(RRe) Then

From Figure 425 we dedu will come from radii at we mation to evaluate the int

When equation (439) is u

The weak dependence up correlation between Dn an If we adopt a distance of relation becomes

D

kr

with a 15 scatter from g

Dwarf elliptical galaxie brightness profiles of giant

15 Recall that (I)eJLB oc -